I have spent a lot of the last twenty-five years giving design and technology work to nonprofits, co-ops, and arts groups in Minneapolis. This is the question every one of them asks now, usually quietly.
Where does AI actually help a nonprofit?
In the back office, where the hours go and nobody is watching.
- Grant writing. Not writing the grant. Drafting the fourth version of the same organisational history, pulling your numbers into the funder's format, and checking the application against the requirements list. A development person gets a day a week back.
- Donor and member data. Finding the duplicates, the dead emails, the lapsed members who used to give every year. Most small nonprofits have a database nobody trusts. A tool can clean it in an afternoon.
- The inbox and the newsletter. Drafts of the replies that are the same every week. A first draft of the newsletter from the three things that happened this month. A person still reads and sends.
- Board and funder reports. The numbers are already somewhere. The tool assembles the first draft. The director spends the time on what the numbers mean.
Where should it stay out?
Anywhere the work is a person being seen by another person. Intake conversations, counselling, case notes that a human needs to feel their way through, the thank-you call to a major donor. The moment a client or a donor feels like they are talking to a machine, you have spent trust you cannot buy back. That is the whole robot suit idea: the tool goes on the staff member, it does not replace them at the door.
Will our donors or members mind?
They mind if you hide it or if it makes the organisation feel cheaper. They do not mind that the newsletter got drafted faster. A one-line policy on the website, "we use AI tools for drafting and data work, never for decisions about the people we serve," handles most of it. I write one with every nonprofit workshop.
What does it cost to start?
A half day with your staff, on paper, finding the hours is $1,500, and nonprofits, co-ops, and arts groups get a rate below that. You leave with a list of what to fix first and a one-page policy. Most of the first wins use tools you already pay for and have never turned on.
What about our data and privacy?
Treat it like a filing cabinet. Client names, health information, and anything a funder would be upset to see in the wrong place never go into a public tool. Paid business accounts from the major vendors do not train on your data; the free ones often do. The policy says which is which, in plain English, so a volunteer can follow it.
Answered by Jared Lukes, Minneapolis. Twenty-five years in brand, UX, and technology, currently holding fractional AI and technology seats at three companies. About Jared.